Tag Archives: Analysis

And the point of that was….?

Statistic Exam 2010.

Well this just made NO sense what-so-ever. Yes, okay I admit this post is going to be doused in a mire of bias and bitterness based on my performance in my stats exam today, but you know what? I don’t care, because at least writing it whilst still trapped behind the heavy fog of oh bollocks means this will at least be an accurate record of my feelings towards Stats on this particular day in my life.

I’m sure there have been (and will be) days when I’ve really enjoyed stats, and would have liked nothing more than to run an ANOVA, or a Multiple Regression to see if my hard work had been ‘statistically’ worth it, but today is NOT one of those days. It’s something satisfying to find nice P values on SPSS if you’ve done lots of work, not quite as satisfying as eating cake or anything crazy but a nice accomplished feeling.
When as a researcher would anyone ever find themselves at a computer unable to deduce what statistical test would be best to use and NOT be allowed to have access to any form of book/google to check their choice for analysis?

Well the answer is never, unless I dont know, you were collecting data on peoples reactions to tornados and as you go to analyse your data the same tornado you are studying rips through your study, swoops up your internet wireless hub and all your text book but kindly spares you, your desk, your laptop and electrical connection.. and you have only 2 hours to complete your work.. or else the tornado will kill again. This seems unlikely.

So why do we have these exams? Would it not be better to give us mini assignments throughout the year, to test our skills on SPSS? I know it’s good to know the background of analysis in order to be better able to make appropriate choices for running a particular test and when you have done so better understand what you can claim from results, but why give us what is at best a memory test to demonstrate an ‘understanding’?

We had two hours fifteen minutes to answer four questions with a range of 2 or 4 subquestions in each section. Not one of these had a percentage next to them either to indicate the worth of the question like our past paper/sample paper had. I had no idea how much to write for any question?!

I attempted all the questions, I definitely knew the answers to about 5 of them out of about 15, (the part a’s usually which I guess are only worth 5% each) and I definitely made up the answer to 3 of the questions (probably the 3 questions worth 20%!) But hey ho. I ended up leaving an hour and 15 minutes early… So I stayed for a whole hour.

1/4 of that hour was spent in the ‘reading time’ and I spent 2/4 of the hour writing, and 1/4 proof reading and drawing a butterfly..

Hope others have experienced a pointless feeling in an exam before. Whilst I know all of the evidence presented above suggests I’ve done badly in this paper, I also can not find it in me to feel disappointed about this. It could be the beta blockers I’m on preventing me from feeling ‘panic’ or it might just be my sense of perspective about this whole issue, statistics are not significant to me.


Leave a comment

Filed under Exams

fMRI is not a ‘Brain Telescope’, yet.

The telescope was and continues to be one of the most important inventions in the study of astronomy, it led to the discovery and understanding of our entire solar system and continues to add to our knowledge of the universe!  This is because it is understood, and the methods used in the application of the telescope are designed to further our knowledge to the best of its ability. It improves over time, and in doing so allows our knowledge and understanding of some of the most unfathomable ideas about our universe to improve also. Neuroscience is not lucky enough to have a brain telescope, and the most used method of investigation in current times fMRI is far from comparable to the telescope, if anything it is more like a kaleidoscope.

Great basis for a research tool, not fully understanding it’s connection to what the ‘activity’ shown is interpret reflect and what is actually being recorded.  I suppose this is why some people call fMRI ‘the new phrenology‘?

fMRI could be equated to using a hammer to put a screw in the wall, sort of does the job but it doesn’t have much of a hold once it’s in there. There must be a better analogy out there but I just can’t think of it, perhaps it’s because if someone else or another scientific field tried to apply the same certainty to the results of a method it was using without fully understanding it they’d be laughed at. It can’t be all bad though surely, otherwise we wouldn’t be using this incredible machine for such rigorous academic investigations?

So I reckon that the general opinion people have of fMRI data is that a brightly coloured area on the brain shows an increase in some unqualified sort of ‘activity’. In fact this is exactly how it’s often reported in published (and therefore peer-reviewed) articles. It’s not just the lay persons or the humble students view, that the ‘areas that light up’ in these scans are meant to be showing which brain areas are involved in what ever process is being studied, but its more surprising that often in published work this assumption is taken for granted. What do I mean? Well yes, good question, let me explain.

Firstly, I should say that what those ‘coloured’ images are actually showing is something call the BOLD signal. This is the blood oxygen level dependent signal. It reflects changes in blood oxygen levels in particular brain regions. This is important to remember because seeing blood oxygen levels change in one area is not the same as seeing neural activity in that area. An important fact often ignored.

Not wanting to get too scientific about this, but not having much choice, what we need to consider is something called ‘neurovascular coupling‘. This refers to how neural activity relates to respiration. Neural activation of any kind requires energy, this energy comes primarily from glucose carried in cerebral blood flow, though it has been suggested that other substances such as glutamate, lactate and glutamine as well as some others, may also be oxidised by neurons (Zielke et al., 2007 in Mangia et al, 2009). Many studies have been conducted using laboratory animals that show a conclusive link between glucose consumption and localised neural activity, though the distribution of that energy consumption is far less straight forward amongst individual neural mechanisms. A study on the brains of rodents suggest that the majority of energy, up to 34% of all being used, is actually used in postsynaptic excitatory neural responses as opposed to an even distribution of  resources involving presynaptic responses, synaptic terminals and maintaining the resting potentials of the cells. In humans this proportional distribution of energy for postsynaptic events could be as high as 74% (Attwell & Laughlin, 2001). It could be that the BOLD signal is therefore most correlated to postsynaptic activity within the brain.

Sounds nice and convincing right? Well no, cause it’s also been suggested that BOLD signal might be showing presynaptic activity, particularly in the glia cells which use vast amounts of energy consumption in their active state (Jueptner & Weiller, 1995). It might seem inconsequential to worry about which bit of neural activity fMRI is showing, the beginning or the end but this is of great importance for furthering our understanding of what is actually happening.

It is also worth remembering that there is a known delay in haemodynamic response to neural activity. That is it takes a while for blood to reach an area where energy is depleted and respiration needs to occur. One way of countering this is using a particular form of analysis when looking at the raw data. Methods such as Canonical Correlation Analysis (Friman et al. 2001) not only remove a lot of the noise from fMRI data samples but also takes into account haemodynamic response delay. Good huh?

I think that it is unfair to label fMRI the new phrenology, it is by far one of the most impressive inventions of recent times and has the potential to inform us about some of the mysteries surrounding what it is to be human and how our magnificent brains work. Give it time, like the telescope, it’s design and application just need to be improved, and this takes centuries.

One day it might just reveal to us the question that gives the answer to ‘the meaning of life, the universe and everything’, if Douglas Adams was correct.

I strongly believe that the best way forward for the moment, given the lack of understanding of mechanisms such as neurovascular coupling is that fMRI analysis continues but is treated with far more caution, and that perhaps combination methods such as MEG and fMRI, or EEG and fMRI might be able to provide more accurate results about the levels and areas of activity.

Don’t always trust those pretty pictures you see, just because they look nice.

P.s and while were at it can we just call MRI by its correct acronym?? NMRI.. that would be NUCLEAR magnetic resonance imaging.  Why do we pander to the concerns a simple and correct use of that word will have on public opinion?

This is science, deal with it.

1 Comment

Filed under Sciencey